&get_box_var;

ORIGINAL ARTICLE

Acute Outcomes and 1-Year Mortality of Intensive Care

Unitacquired Weakness
A Cohort Study and Propensity-matched Analysis
Greet Hermans1,2, Helena Van Mechelen2, Beatrix Clerckx2,3, Tine Vanhullebusch2, Dieter Mesotten2,3,
Alexander Wilmer1, Michael P. Casaer2,3, Philippe Meersseman1, Yves Debaveye2,3, Sophie Van Cromphaut2,3,
Pieter J. Wouters2,3, Rik Gosselink4, and Greet Van den Berghe2,3
1
Medical Intensive Care Unit, Department of General Internal Medicine, and 3Department of Intensive Care Medicine, University Hospitals
Leuven, Leuven, Belgium; and 2Laboratory of Intensive Care Medicine, Division of Cellular and Molecular Medicine, and 4Department of
Rehabilitation Sciences, KU Leuven, Leuven, Belgium

Abstract
Rationale: Intensive care unit (ICU)-acquired weakness is a frequent
complication of critical illness. It is unclear whether it is a marker or
mediator of poor outcomes.
Objectives: To determine acute outcomes, 1-year mortality, and
costs of ICU-acquired weakness among long-stay (>8 d) ICU
patients and to assess the impact of recovery of weakness at ICU
discharge.
Methods: Data were prospectively collected during a randomized

controlled trial. Impact of weakness on outcomes and costs was

analyzed with a one-to-one propensity-score-matching for baseline
characteristics, illness severity, and risk factor exposure before
assessment. Among weak patients, impact of persistent weakness at
ICU discharge on risk of death after 1 year was examined with
multivariable Cox proportional hazards analysis.
Measurements and Main Results: A total of 78.6% were admitted
to the surgical ICU; 227 of 415 (55%) long-stay assessable ICU
patients were weak; 122 weak patients were matched to 122 not-weak

patients. As compared with matched not-weak patients, weak

patients had a lower likelihood for live weaning from mechanical
ventilation (hazard ratio [HR], 0.709 [0.5490.888]; P = 0.009),
live ICU (HR, 0.698 [0.5530.861]; P = 0.008) and hospital
discharge (HR, 0.680 [0.5140.871]; P = 0.007). In-hospital costs
per patient (130.5%, 15,443 Euro per patient; P = 0.04) and 1-year
mortality (30.6% vs. 17.2%; P = 0.015) were also higher. The 105
of 227 (46%) weak patients not matchable to not-weak patients
had even worse prognosis and higher costs. The 1-year risk of
death was further increased if weakness persisted and was more
severe as compared with recovery of weakness at ICU discharge
(P , 0.001).
Conclusions: After careful matching the data suggest that ICU-

acquired weakness worsens acute morbidity and increases

healthcare-related costs and 1-year mortality. Persistence and
severity of weakness at ICU discharge further increased 1-year
mortality.
Clinical trial registered with www.clinicaltrials.gov (NCT 00512122).
Keywords: muscle weakness; costs and cost analysis; mortality;

critical illness; muscle strength

( Received in original form December 23, 2013; accepted in final form May 11, 2014 )
Supported by the Research Foundation-Flanders, Belgium (G.0399.12 and G.0592.12). G.H. received a Postdoctoral Fellowship from the Clinical
Research Fund of the University Hospitals Leuven, Belgium. M.P.C. received a Doctoral Fellowship, and D.M. received a Fundamental Clinical Research
Fellowship of the Research Foundation-Flanders. G.V.d.B., via the University of Leuven, receives structural research financing via the Methusalem program,
funded by the Flemish Government (METH08/07) and holds an ERC Advanced grant (AdvG-2012-321670) from the Ideas Program of the EU FP7.
Author Contributions: G.H., R.G., and G.V.d.B. designed the study. G.H., H.V.M., B.C., T.V., D.M., A.W., M.P.C., P.M., Y.D., S.V.C., and P.J.W. acquired
the data. The statistical analyses were done and interpreted by G.H., H.V.M., and G.V.d.B. G.H. and G.V.d.B. wrote the paper, which was critically
reviewed for important intellectual content by all authors.
Correspondence and requests for reprints should be addressed to Greet Van den Berghe, M.D., Ph.D., Clinical Department and Laboratory of Intensive Care
Medicine, KU Leuven, Herestraat 49, B-3000 Leuven, Belgium. E-mail: greet.vandenberghe@med.kuleuven.be
This article has an online supplement, which is accessible from this issues table of contents at www.atsjournals.org
Am J Respir Crit Care Med Vol 190, Iss 4, pp 410420, Aug 15, 2014
Copyright 2014 by the American Thoracic Society
Originally Published in Press as DOI: 10.1164/rccm.201312-2257OC on May 13, 2014
Internet address: www.atsjournals.org

410

American Journal of Respiratory and Critical Care Medicine Volume 190 Number 4 | August 15 2014

ORIGINAL ARTICLE

At a Glance Commentary
Scientic Knowledge on the
Subject: Clinical weakness occurs

frequently in prolonged critically ill

patients. It remains controversial
whether it is a marker or a mediator of
poor outcomes.
What This Study Adds to the
Field: After accounting for the

potential confounding effects of other

risk factors, it was shown that intensive
care unit (ICU)-acquired weakness
related to delayed weaning from
mechanical ventilation, extended ICU
and hospital stays, more healthcarerelated hospital costs, and a higher risk
of death at 1 year after ICU admission.
These data support causality of the
association between weakness and
poor acute morbidity outcomes and,
even more importantly, late death. The
data underscore the importance of
identifying strategies to prevent and
treat this debilitating problem and
suggest closer follow-up of patients
with ICU-acquired weakness, also after
hospital discharge.
Intensive care unit (ICU)-acquired muscle
weakness (further referred to as weakness) is
a frequent complication of critical
illness. Patients with sepsis, multiple
organ failure, or prolonged mechanical
ventilation in particular are susceptible
for development of weakness, which
affects limbs and respiratory muscles (1).
Given the strong association between
weakness and multiple organ failure,
weakness was considered to be the failure
of just another organ in critically ill patients
(2, 3). From this perspective and not
surprisingly, observational studies revealed
strong associations between weakness
and poor prognosis (412), although this
was not consistently conrmed (13, 14).
Inferentially, weakness may generate
more healthcare-related costs (15), but this
has not been specically investigated. It
remains unclear to what extent weakness is
more a marker than a mediator of poor
outcomes. Also, it is unknown whether the
degree of recovery from weakness at ICU
discharge has any impact beyond the time
of the index hospitalization, such as 1-year
mortality. To address these questions, we

investigated a large cohort of patients

who were systematically screened for
weakness after at least 1 week in ICU.
We compared weak and not-weak patients
after matching for potential confounders,
using propensity score matching. This
method is frequently used in observational
studies to estimate effects if randomization
is not feasible. It is more effective to
reduce bias than multivariate regression
analysis (16) and allows including a larger
number of potentially confounding
variables in the analysis (17). In this
study we rst described the characteristics
of prolonged critically ill patients who
developed weakness after an ICU stay of
at least 8 days. We then evaluated the
impact of weakness on short-term
outcomes, healthcare-related in-hospital
costs, and 1-year mortality. Finally, we
assessed whether persistence of weakness
at ICU discharge and the severity thereof
had any impact on mortality 1 year
after ICU admission. Some of these
results have been previously reported in the
form of an abstract to the International
Symposium on Intensive Care and
Emergency Medicine, 2014 (18).

Methods
Patients and Diagnosis of Weakness

This study was a prospectively planned

subanalysis of the EPaNIC trial (19, 20).
From December 2008 onward, patients
still in ICU on Day 8 after admission
(referred to as long-stay patients) were
systematically assessed for awakening
and cooperation based on the response to
ve commands (4, 21). When adequate
response to all of these was present,
patients were evaluated for weakness by
one of two trained physiotherapists. This
evaluation was repeated three times a
week until ICU discharge or death (22).
Weakness was diagnosed when the Medical
Research Council (MRC) sum score was
less than 48 (21, 22).
At the same time points, inspiratory
muscle strength was measured using
a maximal volitional maneuver, excluding
patients with an articial airway (see
online supplement). The study protocol
and informed consent forms were approved
by the Leuven University Hospital Ethics
Committee (ML4190). All patients
received progressive and systematic
passive and active mobilization. Further

details are described in the online

supplement.
Data Collection and Endpoints

Baseline characteristics were collected

on admission. Baseline risk factors for
development of weakness and other
studied outcomes were age, Acute
Physiology and Chronic Health Evaluation
(APACHE) II score, sex, body mass
index (BMI), nutritional risk score
(23), diabetes mellitus, malignancy,
preadmission dialysis, sepsis, chronic
obstructive pulmonary disease, admission
category, and randomization group
(early or late parenteral nutrition). Risk
factors during ICU stay up to the time of
rst MRC sum score evaluation included
treatment with corticosteroids and
neuromuscular blocking agents, mean
morning blood glucose, occurrence
of new infections, and time to rst MRC,
reecting the time to awakening.
These were previously described (22).
Further details are provided in the online
supplement.
The acute primary endpoint was timeto-live hospital discharge; the medium-term
primary endpoint was mortality 1 year
after ICU admission. Secondary outcomes
were time-to-live ICU discharge, timeto-live weaning from mechanical
ventilation, inspiratory muscle strength,
total billed healthcare costs per patient, ICU
and hospital mortality rates, 6-minute-walk
distance (6MWD) at hospital discharge,
and categories of the healthcare-related
billed costs. 6MWD at hospital discharge
was analyzed on available data and after
imputation of 0 m for patients unable
to walk but not for nonsurvivors because we
aimed to evaluate the functional impact of
weakness at hospital discharge, independent
of any possible mortality effect. Total
healthcare-related costs billed to the health
insurance and the patient were retrieved
from the patients invoices (24). Costs
were divided into Period 1, covering ICU
admission to ICU discharge, and Period 2,
extending from ICU discharge until
hospital discharge. Eight cost categories
were explored: (1) fees (for medical and
allied healthcare-related services), (2)
pharmacy, (3) hospitalization costs per
diem, (4) blood products and other uids,
(5) clinical chemistry, (6) radiology, (7)
graft products (vascular grafts, mechanical
valves, skin grafts, locomotor grafts, and
so forth), and (8) miscellaneous.

Hermans, Van Mechelen, Clerckx, et al.: Acute Outcomes and 1-Year Mortality of ICUAW

411

ORIGINAL ARTICLE
The 1-year mortality was determined
via the national registry for Belgian
citizens and via direct contact with
patient or relatives for foreigners. In
a further exploratory and therefore
inevitably retrospective analysis, we
recorded the destination at hospital
discharge and the details of ICU
and hospital deaths (see online
supplement).

Statistical Analyses

Descriptive statistics included median

and interquartile ranges for continuous
variables and numbers and percentages
for categorical variables. Results were
compared with Mann-Whitney U test and
chi-square test as appropriate.
To examine the impact of the presence
or absence of weakness among long-stay
patients on outcome and healthcare-related

costs, we selected a subset of patients with

and without weakness matched for baseline
risk factors and other risk factors that
occurred during ICU stay up to the time
of rst MRC measurement and known to
be associated with weakness or overall
outcome (Table 1). Matching was based on
propensity scores obtained by logistic
regression and using one-to-one nearest
neighbor matching without replacement

Table 1. Baseline Characteristics and Risk Factors for Weakness in the Total Long-Stay Population, Matched Population, and
Unmatched Weak Patients
Total Population

Baseline characteristics
Age, yr, median (IQR)
APACHE II score,
median (IQR)
Male sex, n (%)
BMI , 25 or . 40, n (%)
NRS , 5, n (%)
Diabetes mellitus, n (%)
Malignancy, n (%)
Preadmission dialysis,
n (%)
Sepsis, n (%)
COPD, n (%)
Admission category
Abdominal/pelvic
surgery, n (%)
Cardiac surgery, n (%)
Cardiovascular, n (%)
Gastrointestinal/
hepatic, n (%)
Hematologic/oncologic,
n (%)
Neurologic, n (%)
Neurosurgery, n (%)
Renal, n (%)
Respiratory, n (%)
Thoracic surgery, n (%)
Transplant, n (%)
Trauma/burns, n (%)
Vascular surgery, n (%)
Other, n (%)
Randomization, late PN,
n (%)
Risk factors occurring during
Time to rst MRC, d,
median (IQR)
Corticosteroids, d,
median (IQR)
NMBA, yes, n (%)
Mean morning glycaemia,
mg/dl, median (IQR)
New infection, n (%)

Matched Population

Unmatched Population

Weak
(n = 227)

Not Weak
(n = 188)

P Value

Weak
(n = 122)

Not Weak
(n = 122)

Standardized
Mean
Difference

64 (5673)
35 (2940)

61 (5074)
31 (2337)

0.097
<0.001

64 (5473)
33 (2539)

65 (5475)
34 (2737)

20.018
20.015

65 (5773)
37 (3342)

72
64
81
21
35
1

(59)
(52.5)
(66.4)
(17.2)
(28.7)
(0.8)

20.016
20.033
0.017
0.023
0
0

56
68
54
13
30
3

127
130
136
35
65
4

71
62
82
22
35
1

136 (59.9)
44 (19.4)

94 (50.0)
42 (22.3)

0.043
0.459

69 (56.6)
24 (19.7)

63 (51.6)
23 (18.9)

36 (15.9)

18 (9.6)

0.089

16 (13.1)

13 (10.7)

63 (27.8)
1 (0.4)
16 (7.0)

53 (28.2)
1 (0.5)
12 (6.4)

36 (29.5)
0 (0)
11 (9)

38 (31.1)
1 (0.8)
11 (9)

27 (25.7)
1 (1.0)
5 (4.8)

9 (4.0)

1 (0.5)

2 (1.6)

1 (0.8)

7 (6.7)

(0.0)
(0.5)
(1.6)
(6.4)
(8.5)
(12.2)
(13.3)
(4.8)
(7.4)
(52.1)

0.200

9 (812)

<0.001

3 (010)

0 (06)

131 (57.7)
102 (96109)

61 (32.4)
103 (98110)

159 (70.0)

101 (53.7)

ICU stay
12 (920)

0
1
3
12
16
23
25
9
14
98

(1.6)
(0)
(0.8)
(4.9)
(11.5)
(9)
(5.7)
(5.7)
(7.4)
(49.2)

0.462
0.034
0.016
0.240
0.984
0.245

0.100
0.021

67 (63.8)
20 (19.0)

0.266
0.906

0.040

20 (19)

0.070

0.033

0
1
2
13
6
10
4
1
8
44

11 (915)

11 (814)

0.019

16 (1123)

<0.001

<0.001

1 (08)

0 (06)

0.098

8 (015)

<0.001

<0.001
0.521

56 (45.9)
103 (96110)

50 (41)
103 (97108)

75 (71.4)
101 (97108)

<0.001
0.694

78 (63.9)

0
1
3
7
12
13
5
5
12
58

(53.3)
(64.8)
(51.4)
(12.4)
(28.6)
(2.9)

(0)
(0.8)
(2.5)
(5.7)
(9.8)
(10.7)
(4.1)
(4.1)
(9.8)
(47.5)

0.001

2
0
1
6
14
11
7
7
9
60

(58.2)
(50.8)
(67.2)
(18)
(28.7)
(0.8)

0.482
<0.001

0.103
0.206
0.004
0.882
0.406
0.253

(0.9)
(0.4)
(1.3)
(8.4)
(8.8)
(9.3)
(4.8)
(3.5)
(7.5)
(45.8)

120
96
138
28
47
1

P Value*

(63.8)
(51.1)
(73.4)
(14.9)
(25.0)
(0.5)

2
1
3
19
20
21
11
8
17
104

(55.9)
(57.3)
(59.9)
(15.4)
(28.6)
(1.8)

Weak
(n = 105)

78 (63.9)

0.099
20.013
0

(0)
(1)
(1.9)
(12.4)
(5.7)
(9.5)
(3.8)
(1)
(7.6)
(41.9)

81 (77.1)

0.273

0.030

Definition of abbreviations: APACHE II = Acute Physiology and Chronic Health Evaluation; BMI = body mass index; COPD = chronic obstructive pulmonary
disease; ICU = intensive care unit; IQR = interquartile range; NMBA = neuromuscular blocking agents; NRS = Nutritional Risk Score; PN = parenteral
nutrition.
Bold values indicate P , 0.05.
*P value for not-matched versus matched weak patients.

All risk factors were calculated up to the time of first Medical Research Council sum score evaluation for every patient individually.

412

American Journal of Respiratory and Critical Care Medicine Volume 190 Number 4 | August 15 2014

ORIGINAL ARTICLE
with weakness as the dependent variable
(25). To optimize matching for all variables
of interest, time-to-rst MRC was entered
as a log10 transformed factor. The caliper
was gradually narrowed, starting from
0.2, to obtain satisfactory matching as
indicated by an absolute standardized
difference in means less than or equal to
0.1 for all variables. The standardized
mean difference was dened as the mean
difference between the groups divided by
the standard deviation of the control group
(26). This was reached at a caliper of 0.1
(i.e., 0.1 3 standard deviation of the logit of
the propensity score). For time-to-event
analyses, comparisons for patients with
and without weakness were done with
Cox proportional hazards analysis and
visualized with Kaplan-Meier plots. Because
the time-to-event analyses were performed
in a subgroup matched for confounding
factors, no additional adjustments for these
were made in the Cox regression model.
Time-to-alive weaning was calculated from
ICU admission. Time-to-alive ICU and
hospital discharge were calculated from the
time of measurement of MRC sum score.
A robust estimator of variance was used for
analyses of paired data (27). To further
assess the impact of persistence and
severity of weakness at ICU discharge on
medium-term prognosis, we analyzed the
association between weakness at ICU
discharge and 1-year survival among
weak patients with multivariable Cox
proportional hazards analysis. Patients were
categorized as recovered from weakness
(MRC sum score > 48) or with persisting
weakness (with either 48 . MRC sum score
> 36, or MRC sum score , 36). Analysis
was performed with a forward stepwise
method (likelihood ratio, probability for
enter 0.05, removal 0.1), including all
baseline risk factors potentially affecting
survival, and the risk factors to which the
weak patients were exposed before
diagnosis of weakness and that were
potentially related with survival. For this
purpose, because limited confounders can
be included in multivariable models, the
16 admission categories were grouped into
four main categories for this analysis,
as described (see Table E1 in online
supplement) (22). This analysis was
performed on the total population of weak
patients, because we expected that the
matched subset would be less severely ill
and not completely representative for all
weak patients. The time variable entered in

the model was calculated from the last

MRC measurement up to 1 year after ICU
admission.
All analyses were performed with
IBM SPSS-20 (IBM, Armonk, NY).
Propensity score matching was performed
with IBM SPSS-20 and R version R2.10.1
(R Foundation for Statistical
Computing, Vienna, Austria) (26).
Differences were considered signicant
when two-sided P values were 0.05 or
less.

Results
Patient Characteristics

Between October 2008 and November 2010

MRC sum score was measured in 415
long-stay ICU patients (Figure 1). The
population constituted of 28% admissions
following cardiac surgery; 47.2% urgent
admissions for complications after other
surgery, burns, and trauma; 3.4% elective
admissions following other surgery; and
21.4% admissions to the medical ICU.
Weakness was present in 227 (55%)
patients. Baseline characteristics and

exposure to risk factors for weakness during

ICU stay and up to the moment of actual
measurement of MRC sum score are listed
in Table 1. Weak patients were more
severely ill on admission than not-weak
patients as reected by the APACHE II
score (35 [2940] vs. 31 [2337]), less often
had a low nutritional risk score (59.9%
vs. 73.4%), and more often had sepsis on
admission (59.9% vs. 50.0%). Weak patients
were treated more often and longer with
corticosteroids (3 d [010] vs. 0 d [06]),
more frequently received neuromuscular
blocking agents (57.7% vs. 32.4%), and
experienced new infectious episodes
between admission and MRC sum score
evaluation (70% vs. 53.7%). Time-toawakening and rst MRC sum score
measurement was signicantly longer
in weak than in not-weak patients (12 d
[920] vs. 9 d [812]). Mortality rates of the
studied patients were relatively low despite
high severity of illness. This is caused by
the selection of long-stay patients who were
awake and fully cooperative in ICU. By this
selection, we omitted severely ill patients
who died early in ICU and long-stay
patients who were not awake and

Figure 1. Flow chart of patients evaluated. ICU = intensive care unit; ICUAW = intensive care
unitacquired weakness; NMD = neuromuscular disease; MRC = Medical Research Council.

Hermans, Van Mechelen, Clerckx, et al.: Acute Outcomes and 1-Year Mortality of ICUAW

413

ORIGINAL ARTICLE
cooperative enough for testing and who
clearly have worse outcomes than those
who were studied (see Table E2). This
selection also explained the substantial
difference between ICU and hospital
mortality for the studied population (data
not shown).
Impact of Presence of Weakness in
Long-Stay Patients on Outcomes and
Healthcare-related Costs

Before matching and as compared with

patients without weakness, weakness was
associated with poorer acute outcomes. At
any time, patients with weakness had a
signicantly lower likelihood of being alive
and weaned, discharged from ICU and from
the hospital than patients without weakness.
Details on respiratory muscle strength are
reported in Table E3. Weak patients had
a higher ICU and hospital mortality
(Table 2). The circumstances of these
deaths were further analyzed retrospectively
(see Table E4). Although statistically
more weak patients were DNR coded at the
time of death (P = 0.044), most patients
died while care was withdrawn. No
signicant difference was detected in the
incidence of readmissions, recurrence of
respiratory failure, possible aspiration,
tracheostomy, or cause of death. These
data are limited by their retrospective
nature and by the low statistical power
because only 12 deaths occurred in notweak patients. In fewer weak as compared
with not-weak patients, 6MWD could be
obtained at hospital discharge. Reasons
for not being walked for 6 minutes
varied for weak and not-weak patients.
When tested, the distance walked in 6
minutes did not signicantly differ.
After imputation of 0 m for those
patients who were unable to walk for
physical or mental reasons, weak
patients walked signicantly less distance
in 6 minutes. Similar results were
obtained when also imputing 0 for
nonsurvivors. Discharge destination was
signicantly different for weak versus
not-weak patients. Weakness was
associated with more incremental
healthcare-related costs and higher 1-year
mortality.
Because large imbalances in baseline
characteristics and in other risk factors
for the development of weakness were
found between patient groups, the actual
contribution of weakness to worse outcomes
and increased healthcare costs, independent
414

from other covariates, was examined in

a matched subset of patients. This
propensity score based matching procedure
resulted in 122 unique pairs of patients
with and without weakness, who were well
matched for baseline characteristics and
known risk factors for weakness to
which they were exposed before the
measurement of MRC sum score (Figure 2,
Table 1). In this matched population,
and as compared with not-weak patients,
patients with weakness at any time had
a signicantly lower likelihood for being
alive and weaned from mechanical
ventilation (hazard ratio [HR], 0.709
[0.5490.888]; P = 0.009) (Figure 3A), for
being alive and discharged from the ICU
(HR, 0.698 [0.5530.861]; P = 0.008)
(Figure 3B), and from the hospital (HR,
0.680 [0.5140.871]; P = 0.007)
(Figure 3C). In 14.8% of weak patients
and in 27.9% of not-weak patients,
6MWD was obtained at hospital
discharge (P = 0.012). Reasons for missing
6MWD data did not signicantly differ
between groups, and also distance walked
within 6 minutes when tested did not
differ. However, when a 0 value was
imputed for patients with new physical or
mental impairment precluding
evaluation, the 6MWD distance was
signicantly lower (P = 0.01). Discharge
destination was signicantly different for
weak versus not-weak patients (P = 0.017)
with, respectively, 17.5% versus 9.8%
of the patients being discharged to
rehabilitation units and 18.4 versus 8.9%
to other hospitals. ICU mortality (P =
0.355) and hospital mortality rate (P =
0.075) were not different, but mortality
after 1 year was higher in weak than in
not-weak patients (30.4% vs. 17.2%; P =
0.015). This effect remained when
matching procedure was repeated with
additional separation of BMI less than
25 and BMI greater than 40 (see online
supplement).
Also after matching, total billed costs
for the hospitalization remained higher
in weak than in not-weak patients with
a median difference of 5,443 Euros
(130.5%; P = 0.04). When dividing costs
for the rst ICU and second ward period,
only costs for the ICU period were
signicantly higher (13,794 Euros per
patient or 127.9%; P = 0.048). The
differences for this period were mainly
attributable to costs for clinical chemistry,
radiology, and graft products, but the

latter was of no relevance because the

median cost for this category was 0 Euros
per patient for both groups (see Table E5).
Costs of clinical chemistry (R2 = 0.886)
and radiology (R2 = 0.778) were strongly
related with duration of ICU stay. In
bivariate analysis, the longer duration of
ICU stay for weak as compared with notweak patients explained the differences
for clinical chemistry. Indeed, by adding the
duration of ICU stay to the model, the
independent association with weakness
was lost (P = 0.214) and was taken over
by that with duration of ICU stay (P ,
0.001). For radiology, both presence
of weakness and (P = 0.03) duration of ICU
stay (P , 0.001) remained independently
associated.
Notably, 105 of 227 (46%) weak
patients could not be matched to patients
without weakness (Table 1). These
unmatched and weak long-stay patients
were sicker at admission, with higher
APACHE II scores, and more frequently
had a high nutritional risk score
(nutritional risk score .5) and low or
high BMI, as compared with the weak
patients who did get matched. The
unmatched group had signicantly
more exposure to known risk factors for
weakness before assessment, such as
corticosteroids and neuromuscular
blocking agents, as compared with the
weak but matched patients (Table 1). Timeto-awakening and rst MRC sum score
measurement was also signicantly higher
and not-matched patients more often
developed new infections in ICU before
MRC sum score evaluation. The notmatched weak patients had signicantly
worse outcomes than the matched weak
patients with a median increase of time-tolive weaning of 9 days, time-to-live ICU
discharge of 4 days, and time-to-live
hospital discharge of 16 days. Total billed
cost for this unmatched subgroup of weak
patients was median 8,057 Euros higher
(135%; P , 0.001) than in the matched
weak patients.
Impact of Recovery of Weakness at
ICU Discharge on 1-Year Mortality

Among the 227 weak long-stay patients, risk

of death at 1 year after ICU admission
was dependent on the persistence of
weakness at ICU discharge and on severity
of such persistent weakness (P , 0.001)
(Figure 4). At any time within the rst year
following ICU admission, compared with

American Journal of Respiratory and Critical Care Medicine Volume 190 Number 4 | August 15 2014

Hermans, Van Mechelen, Clerckx, et al.: Acute Outcomes and 1-Year Mortality of ICUAW
7 (412)
3 (16)
5 (2.7)

14 (830)

7 (319)

31 (23)

32 (16.8)

72 (31.9)

26,348 (16,63744,519)
19,678 (12,18633,901)
3,633 (1,1438,597)
27 (14.4)

17,356 (11,50730,205)
12,517 (7,69220,523)
2,712 (1,1276,886)

142 (80.7)
17 (9.7)
17 (9.7)

78 (57.8)

86 (45)

114 (63)
39 (21.5)
28 (15.5)

12 (8.9)
14 (10.4)

46 (24.1)
27 (14.1)

244 (185300)
200 (104287)

12 (6.4)
53 (28.2)

46 (20.3)
36 (15.9)

223 (120280)
78 (0240)

22 (1341)

43 (21114)

18 (7.9)

52 (4956)
0 (0)

42 (3444)
69 (30.4)

Not Weak
(n = 188)

Total Population

<0.001

<0.001
<0.001
0.148

0.001

0.102
0.002

19 (15.6)
18 (14.8)

<0.001
0.002
0.002

37 (30.6)

21 (17.2)

17,834 (12,22731,306)
13,622 (8,53920,847)
2,904 (1,0956,911)

91 (81.2)
11 (9.8)
10 (8.9)

214 (163286)
191 (90270)

20 (22.7)

49 (55.7)

10 (11.4)
9 (10.2)

10 (8.2)
34 (27.9)

23 (1341)

4 (3.3)

3 (08)

8 (514)

52 (4956)
0 (0)

Not Weak
(n = 122)

Matched Population

23,277 (15,37036,147)
17,416 (10,08328,470)
3,289 (1,0548,267)

66 (64.1)
18 (17.5)
19 (18.4)

199 (120264)
66 (0207)

20 (19.2)

49 (47.1)

19 (18.3)
16 (15.4)

7 (5.7)
36 (1683)

6 (214)

<0.001
0.02

11 (722)

<0.001

<0.001

42 (3544)
31 (25.4)

<0.001
<0.001

P Value

Weak
(n = 122)

0.015

0.040
0.048
0.675

0.017

0.277
0.010

0.075
0.012
0.328

0.007

0.355

0.008

0.009

<0.001
<0.001

P Value

Weak
(n = 105)

35 (33.3)

31,334 (19,86660,331)
25,539 (15,04850,623)
4,293 (1,3139,258)

48 (61.5)
21 (26.9)
9 (11.5)

239 (105319)
103 (0260)

12 (13.8)

37 (42.5)

27 (31.0)
11 (12.6)

27 (25.7)
18 (17.1)

52 (24312)

11 (10.5)

10 (330)

20 (1141)

0.658

<0.001
<0.001
0.290

0.197

0.406
0.301

0.058
0.623
0.211

0.009

0.188

0.015

0.001

0.077
0.078

P Value*

Unmatched Population

39 (3344)
38 (36.2)

Definition of abbreviations: 6MWD = 6-minute-walk distance; ICU = intensive care unit; MRC = Medical Research Council; MV = mechanical ventilation.
Period 1 covered ICU admission to ICU discharge; Period 2 covered ICU discharge to hospital discharge.
Duration of mechanical ventilation was calculated from ICU admission; ICU and hospital stay were calculated from time of measurement of MRC sum score.
Bold values indicate P , 0.05.
*P value for not-matched versus matched weak patients.

P values for time-to-event analysis were calculated using Cox regression analysis.

Imputation of 0 m was performed for bad outcomes including nonpreexisting physical or mental reasons, which precluded performing 6MWD.

Strength data
First MRC sum score
MRC sum score , 36
ICU stay
Time to alive weaning
from MV, d
Time to alive ICU
discharge, d
ICU mortality, n (%)
Hospital stay
Time to alive hospital
discharge, d
Hospital mortality, n (%)
6MWD performed, n (%)
6MWD, reasons not
performed
Death
Physical or psychological
impairment
Assessments not completed
before discharge
Premorbid limitation/refusal/
assessor not available/not
classiable
6MWD available data, m
6MWD with imputation
data, m
Discharge destination survivors
Home
Rehabilitation unit
Other hospital
Costs
Total billed costs per patient
Period 1
Period 2
Medium-term
One-year mortality, n (%)

Weak
(n = 227)

Table 2. Outcome Characteristics in the Total Long-Stay Population, Matched Population, and Unmatched Weak Patients

ORIGINAL ARTICLE

415

ORIGINAL ARTICLE
Discussion

Figure 2. Mean standardized differences for baseline characteristics, illness severity, and risk
factor exposure before MRC evaluation before and after propensity score matching. The
horizontal axis represents the mean standardized difference, open dots reflect values before
matching, and black dots values after matching. If both values overlap, only the black dot is
visible. Matching procedure aimed at, and succeeded in, reducing mean standardized
difference to an absolute value of maximally 0.1. APACHE II = Acute Physiology and Chronic
Health Evaluation; BMI = body mass index; COPD = chronic obstructive pulmonary disease;
MRC = Medical Research Council; NMBA = neuromuscular blocking agents; NRS = nutritional
risk score.

patients who recovered from weakness

and adjusted for potential confounders,
those with persistent weakness and MRC
sum between 36 and 47 at ICU discharge
had a higher likelihood of death (HR,
2.104; 95% condence interval,

1.1343.903; P = 0.018). This likelihood of

late death was even higher for patients
with a more severe degree of persistent
weakness (MRC sum , 36; HR, 4.273;
95% condence interval, 2.0858.754;
P , 0.001) (Figure 4).

We present a large cohort of long-stay

ICU patients prospectively evaluated for
weakness. Using a one-to-one propensity
score matched analysis, we assessed
impact of weakness on short-term
outcomes, 1-year mortality, and inhospital healthcare-related costs. Weak
patients had worse in-hospital morbidity
(but not mortality outcomes), generated
more hospital costs, and revealed
a higher mortality 1 year after ICU
admission than not-weak patients.
The 1-year mortality of patients who
developed weakness during ICU stay was
further increased when weakness persisted
at ICU discharge, and was even higher
when persistent weakness at ICU
discharge was more severe. This suggests
that ICU-acquired weakness
independently contributes to the legacy of
critical illness.
Neuromuscular complications of
critical illness are common and represent
major functional morbidity (28, 29).
Strategies to prevent weakness are limited
in number and effectiveness (15, 30).
These include aggressive treatment of the
underlying condition, glycemic control
(31, 32), and implementing an early
rehabilitation strategy with minimal
sedation (33, 34). Higher protein delivery in
the rst week was recently associated
with greater muscle wasting (35). Also,
avoiding parenteral nutrition in the rst

Figure 3. Kaplan-Meier plots depicting the proportion of propensity score matched patients over time that were alive and weaned from the ventilator,
discharged from intensive care unit (ICU) and from the hospital. The cumulative proportion of patients weaned alive from mechanical ventilation (A),
discharged alive from the ICU (B), and discharged alive from the hospital (C) are shown for the matched weak and not-weak long-stay patients. Data for
patients who died were censored after the last patient had been weaned alive (A), or discharged alive from the ICU (B) or the hospital (C). Time-to-alive
weaning was calculated from ICU admission. Time-to-alive ICU and hospital discharge were calculated from the time of measurement of Medical
Research Council sum score. ICUAW = intensive care unitacquired weakness.

416

American Journal of Respiratory and Critical Care Medicine Volume 190 Number 4 | August 15 2014

ORIGINAL ARTICLE

Figure 4. Cox regression estimates for survival in the first year after intensive care unit (ICU)
admission in the total population of weak patients according to persistence and severity of weakness
at final examination in the ICU. The survival curve visually displays the model predicted survival
time for the average patient (that is other covariates are fixed at their average values) according to
the Medical Research Council (MRC) sum score at final examination in the ICU: the plot shows
the effect of recovery from weakness, persisting weakness with MRC from 36 to 47, and persisting
weakness with MRC less than 36 by the end of ICU stay. The time variable entered in the model
was calculated from the last MRC measurement up to 1 year after ICU admission.

week in ICU reduced weakness (22).

Several observational studies indicated that
weakness is associated with poor outcomes,
including longer duration of mechanical
ventilation, ICU stay and hospital stay, and
higher ICU and hospital mortality. Others
could not conrm an independent
relationship of neuromuscular complications
in the ICU with outcome (14, 36). This
controversy is at least partially explained by
the difcult clinical diagnosis of weakness
and by the fact that randomized studies to
address the question of causality are not
possible. ICU-acquired weakness could
indeed be just a marker of illness severity
and of poor prognosis. To examine any
potential causal impact of ICU-acquired
weakness on outcome in a mixed population
of long-stay patients, we created 122 unique
pairs of weak and not-weak patients with
similar baseline characteristics and risk
factors for weakness. The population studied
was a subgroup of EPaNIC, a randomized
controlled trial examining the effects of early
versus late parenteral nutrition on overall
outcome. In this trial, early parenteral
nutrition negatively affected muscle
strength, although clearly the nutritional
strategy was unable to fully prevent
weakness (22). For this reason, we included

the randomization arm in the propensity

model. This analysis was performed using
one-to-one propensity score matching
procedure without replacement. Other
methods, such as multiple regression, tend
to inate effects in observational studies,
especially when the number of prognostic
factors is high (16) and when there is
insufcient overlap of covariates between
the two groups of interest (37). By stringent
and conservative matching analysis, we
attempted to get as close as possible to causal
inference of weakness (37).
We found that weak patients had
worse morbidity outcomes than patients
without weakness, as reected by a lower
likelihood at any time for live weaning, ICU
discharge, and hospital discharge. A possible
mechanism explaining worse short-term
outcome is coexistence of respiratory muscle
weakness. Both peripheral and respiratory
muscle weakness are related with severity
of illness and sepsis (3840) and may be the
reection of organ failure. Also, respiratory
muscle weakness is associated with
peripheral muscle weakness (6). A clear
relationship between respiratory muscle
weakness measured using magnetic
stimulation, a method not requiring patient
cooperation, and worse outcome has been

demonstrated (38, 39). Using volitional

measurements of respiratory muscle
strength, we could not conrm reduced
respiratory muscle strength in the matched
population. This may be because of bias
induced by the selection of patients tested
for maximal inspiratory pressure, which did
not allow an articial airway. Therefore,
partial recovery could have been present at
the time of measurement. Also, sample size
reduction with the matching procedure
inevitably further reduced statistical power.
Pharyngeal dysfunction and symptomatic
aspiration, related to limb muscle weakness
in chronically ventilated patients (41), could
be another explanation for the worse
outcome. We cannot conrm this relationship
because we did not systematically assess
swallowing in our patients.
ICU and hospital mortality were not
different. Strikingly, patients who acquired
weakness in the ICU did have higher 1-year
mortality than matched patients without
weakness. Other available data on mediumterm mortality of critically ill patients with
neuromuscular complications are scarce.
Leijten and coworkers (42) reported in
a small subset of 50 severely ill patients with
critical illness polyneuropathy that hospital
mortality was increased, but the sample
size did not have the statistical power to
address signicance of the seemingly higher
1-year mortality (52% vs. 43%).
Our ndings suggest that weakness
diagnosed clinically in ICU affects patients
health beyond ICU and hospital discharge.
This conrms the association between
muscle weakness and impaired physical
function and health-related quality of life in
patients with acute lung injury, shown to
persist up to 24 months after admission
(43). The absence of any signicant impact
of weakness on ICU and hospital mortality
in our population may indicate that the
predominant immediate impact of
weakness is morbidity and delayed recovery
rather than increased risk of death in
the hospital. Alternatively, sample size
reduction by the matching procedure may
have reduced statistical power to detect
differences in ICU and hospital mortality.
Also, the robustness of the statistical
methods we used may explain why an
immediate risk of death was not associated
with weakness, because it was present
before matching. A substantial amount of
long-stay patients (105 of 227) diagnosed
with weakness could not be matched to
patients without weakness, and these

Hermans, Van Mechelen, Clerckx, et al.: Acute Outcomes and 1-Year Mortality of ICUAW

417

ORIGINAL ARTICLE
patients were sicker, had more risk factors
for weakness, and had worse outcomes than
the matched weak patients. Hence, the
propensity-matched analysis represents
a very conservative approach toward the
impact of weakness on outcomes.
With this methodology, increased late
mortality of patients who acquired weakness
during the ICU stay is striking and could
have important implications for patient care.
The shorter distance walked in 6 minutes at
hospital discharge, apparent after imputation
of a poor score for patients unable to walk
for reasons that may mask weakness, as
previously done (44), suggests that the
weakness had functional impact at hospital
discharge. This is further conrmed by the
post hoc analysis of the discharge destination
showing clearly different proportions of
patients being discharged to rehabilitation
units, other hospitals, or home. Our nding
that persistence of weakness at ICU
discharge, and the severity thereof, further
increased the risk of death after 1 year as
compared with patients who were weak but
recovered from weakness before ICU
discharge suggests longer-term consequences
and implications for patient care. Fan and
coworkers (43) recently reported substantial
mortality among survivors of acute lung
injury long after ICU and hospital discharge.
This concurs with the concept that
critical illnessinduced neuromuscular
complications may represent a rapid-onset
frailty across a range of age strata (45), which
itself has been related with increased risk of
adverse events, morbidity, and mortality
(46). Patients diagnosed with weakness after
prolonged ICU stay could possibly benet
from closer follow-up after ICU and hospital
discharge to prevent late death.
Limitations

Results apply to the subgroup of long-stay

but cooperative ICU patients and therefore
cannot be extrapolated to short-stayers or to
long-stayers who never regained enough

cooperation to allow testing. An important

fraction of patients indeed could not be
tested for weakness because they did not
regain adequate awakening at the three
weekly screening moments. Daily screening
could potentially have decreased this
number. We did not use a validated delirium
scale, but requested patients to correctly
respond to ve out of ve complex
commands to avoid testing patients unable
to remain attentive for a sufciently long
period or to perform the complex
commands. We did not measure or adjust
for baseline muscle weakness before ICU
admission because, in general, it is not
feasible to prospectively evaluate this
because of the unplanned nature of critical
care admissions. A comorbidity-derived
measure, such as the Functional
Comorbidity Index (47), designed as
a predictor of physical functioning in ICU
survivors could have been useful for this
purpose. We cannot exclude residual
confounding by this or any other
unmeasured factors. Because of the one-toone propensity score matching procedure
with narrow caliper, the sample size was
reduced. We did not formally address the
causes of late mortality. Therefore, the
mechanisms leading to increased 1-year
mortality remain to be unraveled. Also, the
low percentage of patients that were
evaluated for the 6MWD with a substantial
number of assessments not completed
before discharge may limit conclusions that
can be drawn from these results. To avoid
bias by omitting patients unable to walk
for reasons that may mask weakness, we
imputed 0 values. However, in addition to
weakness, other factors, such as cognitive or
psychiatric complications, gait or balance
disturbances, contractures, or xed exion
of joints from heterotopic ossication,
can limit functionality. Since the
completion of the early exercise training
study in February 2007 performed in some
of the participating ICUs (34), mobilizing

References
1. Stevens RD, Dowdy DW, Michaels RK, Mendez-Tellez PA, Pronovost
PJ, Needham DM. Neuromuscular dysfunction acquired in critical
illness: a systematic review. Intensive Care Med 2007;33:18761891.
2. Hund E. Neurological complications of sepsis: critical illness
polyneuropathy and myopathy. J Neurol 2001;248:929934.
3. Bolton CF. Sepsis and the systemic inammatory response syndrome:
neuromuscular manifestations. Crit Care Med 1996;24:14081416.
4. De Jonghe B, Sharshar T, Lefaucheur JP, Authier FJ, Durand-Zaleski I,
Boussarsar M, Cerf C, Renaud E, Mesrati F, Carlet J, et al.; Groupe de

418

critically ill patients early became standard

of care. Such care was provided in
a protocolized manner (48) and to the best
abilities of the physical therapy team. We
did not actually record number and duration
of sessions. We cannot exclude that optimal
treatment provided in a randomized
setting (33, 34) could have resulted in better
functional outcomes, although the setting
reected well the daily practice. Finally,
although we and others found good
reproducibility of MRC sum score in various
clinical settings (4951), including critically
ill patients (9, 21), others could not
conrm this (14, 36), which may limit
generalizability of conclusions.
Conclusions

Screening for clinical muscle weakness in

patients in ICU for at least 8 days allows
identifying patients with ICU-acquired
weakness that seems to expose them to an
increased risk of short-term morbidity and
a higher risk of late death 1 year after the
acute event. Weakness generated extra
healthcare-related costs predominantly
during the time in ICU rather than on the
regular hospital wards. Our ndings also
stress the importance of further research
aimed at prevention and/or treatment of this
detrimental complication that seems to
contribute to the legacy of critical illness.
These ndings also suggest that weak patients
should be closely monitored following ICU
and hospital discharge to prevent
complications that lead to late death. n
Author disclosures are available with the text
of this article at www.atsjournals.org.
Acknowledgment: The authors are indebted to
all patients and their families for their participation
in this study. The clinical trial assistants, in
particular Alexandra Hendrickx and Sylvia Van
Hulle, and Jenny Gielens are acknowledged for
administrative support. The authors are grateful
to the intensive care unit nursing and medical staff
for excellent patient care.

Reexion

et dEtude des Neuromyopathies en Reanimation.

Paresis
acquired in the intensive care unit: a prospective multicenter study.
JAMA 2002;288:28592867.
5. De Jonghe B, Bastuji-Garin S, Sharshar T, Outin H, Brochard L. Does
ICU-acquired paresis lengthen weaning from mechanical ventilation?
Intensive Care Med 2004;30:11171121.
6. De Jonghe B, Bastuji-Garin S, Durand MC, Malissin I, Rodrigues P, Cerf
C, Outin H, Sharshar T; Groupe de Reexion

et dEtude des
Neuromyopathies en Reanimation.

Respiratory weakness is
associated with limb weakness and delayed weaning in critical illness.
Crit Care Med 2007;35:20072015.

American Journal of Respiratory and Critical Care Medicine Volume 190 Number 4 | August 15 2014

ORIGINAL ARTICLE
7. Sharshar T, Bastuji-Garin S, Stevens RD, Durand MC, Malissin I,
Rodriguez P, Cerf C, Outin H, De Jonghe B; Groupe de Reexion

et
dEtude des Neuromyopathies En Reanimation.

Presence and severity

of intensive care unit-acquired paresis at time of awakening are
associated with increased intensive care unit and hospital mortality.
Crit Care Med 2009;37:30473053.
8. Nanas S, Kritikos K, Angelopoulos E, Siafaka A, Tsikriki S, Poriazi M,
Kanaloupiti D, Kontogeorgi M, Pratikaki M, Zervakis D, et al.
Predisposing factors for critical illness polyneuromyopathy in
a multidisciplinary intensive care unit. Acta Neurol Scand 2008;118:
175181.
9. Ali NA, OBrien JM Jr, Hoffmann SP, Phillips G, Garland A, Finley JC,
Almoosa K, Hejal R, Wolf KM, Lemeshow S, et al.; Midwest Critical
Care Consortium. Acquired weakness, handgrip strength, and
mortality in critically ill patients. Am J Respir Crit Care Med 2008;178:
261268.
10. Brunello AG, Haenggi M, Wigger O, Porta F, Takala J, Jakob SM.
Usefulness of a clinical diagnosis of ICU-acquired paresis to predict
outcome in patients with SIRS and acute respiratory failure. Intensive
Care Med 2010;36:6674.
11. Tzanis G, Vasileiadis I, Zervakis D, Karatzanos E, Dimopoulos S,
Pitsolis T, Tripodaki E, Gerovasili V, Routsi C, Nanas S. Maximum
inspiratory pressure, a surrogate parameter for the assessment of
ICU-acquired weakness. BMC Anesthesiol 2011;11:14.
12. Lee JJ, Waak K, Grosse-Sundrup M, Xue F, Lee J, Chipman D,
Ryan C, Bittner EA, Schmidt U, Eikermann M. Global muscle strength
but not grip strength predicts mortality and length of stay in a general
population in a surgical intensive care unit. Phys Ther 2012;92:
15461555.
13. Latronico N, Bertolini G, Guarneri B, Botteri M, Peli E, Andreoletti S,
Bera P, Luciani D, Nardella A, Vittorielli E, et al. Simplied
electrophysiological evaluation of peripheral nerves in critically ill
patients: the Italian multi-centre CRIMYNE study. Crit Care 2007;11:
R11.
14. Connolly BA, Jones GD, Curtis AA, Murphy PB, Douiri A, Hopkinson
NS, Polkey MI, Moxham J, Hart N. Clinical predictive value of manual
muscle strength testing during critical illness: an observational
cohort study. Crit Care 2013;17:R229.
15. Hermans G, De Jonghe B, Bruyninckx F, Van den Berghe G.
Interventions for preventing critical illness polyneuropathy and
critical illness myopathy. Cochrane Database Syst Rev 2014;1:
CD006832.
16. Martens EP, Pestman WR, de Boer A, Belitser SV, Klungel OH.
Systematic differences in treatment effect estimates between
propensity score methods and logistic regression. Int J Epidemiol
2008;37:11421147.
17. Heinze G, Juni
P. An overview of the objectives of and the approaches
to propensity score analyses. Eur Heart J 2011;32:17041708.
18. Hermans G, Van Mechelen H, Clerckx B, Vanhullebusch T, Mesotten D,
Wilmer A, Casaer MP, Meersseman P, Debaveye Y, Van
Cromphaut S, et al. Acute and long-term outcomes of ICU-acquired
weakness: a cohort study and propensity-matched analysis.
Presented at ISICEM. March 1922, 2014, Brussels, Belgium.
19. Casaer MP, Mesotten D, Hermans G, Wouters PJ, Schetz M, Meyfroidt
G, Van Cromphaut S, Ingels C, Meersseman P, Muller J, et al.
Early versus late parenteral nutrition in critically ill adults. N Engl
J Med 2011;365:506517.
20. Casaer MP, Hermans G, Wilmer A, Van den Berghe G. Impact of early
parenteral nutrition completing enteral nutrition in adult critically ill
patients (EPaNIC trial): a study protocol and statistical analysis
plan for a randomized controlled trial. Trials 2011;12:21.
21. Hermans G, Clerckx B, Vanhullebusch T, Segers J, Vanpee G,
Robbeets C, Casaer MP, Wouters P, Gosselink R, Van Den Berghe
G. Interobserver agreement of Medical Research Council sumscore and handgrip strength in the intensive care unit. Muscle Nerve
2012;45:1825.
22. Hermans G, Casaer MP, Clerckx B, Guiza

F, Vanhullebusch T, Derde S,
Meersseman P, Derese I, Mesotten D, Wouters PJ, et al. Effect
of tolerating macronutrient decit on the development of intensivecare unit acquired weakness: a subanalysis of the EPaNIC trial.
Lancet Respir Med 2013;1:621629.

23. Kondrup J, Allison SP, Elia M, Vellas B, Plauth M; Educational and

Clinical Practice Committee, European Society of Parenteral and
Enteral Nutrition (ESPEN). ESPEN guidelines for nutrition screening
2002. Clin Nutr 2003;22:415421.
24. Vanderheyden S, Casaer MP, Kesteloot K, Simoens S, De Rijdt T,
Peers G, Wouters PJ, Coenegrachts J, Grieten T, Polders K, et al.
Early versus late parenteral nutrition in ICU patients: cost analysis
of the EPaNIC trial. Crit Care 2012;16:R96.
25. Gayat E, Pirracchio R, Resche-Rigon M, Mebazaa A, Mary JY, Porcher
R. Propensity scores in intensive care and anaesthesiology literature:
a systematic review. Intensive Care Med 2010;36:19932003.
26. Thoemmes F. Propensity score matching in SPSS [accessed 2014
May 28]. Available from: http://arxiv.org/ftp/arxiv/papers/1201/
1201.6385.pdf
27. Austin PC. The performance of different propensity score methods for
estimating marginal hazard ratios. Stat Med 2013;32:28372849.
28. Latronico N, Bolton CF. Critical illness polyneuropathy and myopathy:
a major cause of muscle weakness and paralysis. Lancet Neurol
2011;10:931941.
29. Batt J, dos Santos CC, Cameron JI, Herridge MS. Intensive care unitacquired weakness: clinical phenotypes and molecular mechanisms.
Am J Respir Crit Care Med 2013;187:238246.
30. Eikermann M, Latronico N. What is new in prevention of muscle
weakness in critically ill patients? Intensive Care Med 2013;39:
22002203.
31. Van den Berghe G, Schoonheydt K, Becx P, Bruyninckx F, Wouters PJ.
Insulin therapy protects the central and peripheral nervous system
of intensive care patients. Neurology 2005;64:13481353.
32. Hermans G, Wilmer A, Meersseman W, Milants I, Wouters PJ,
Bobbaers H, Bruyninckx F, Van den Berghe G. Impact of
intensive insulin therapy on neuromuscular complications and
ventilator-dependency in MICU. Am J Respir Crit Care Med 2007;
175:480489.
33. Schweickert WD, Pohlman MC, Pohlman AS, Nigos C, Pawlik AJ,
Esbrook CL, Spears L, Miller M, Franczyk M, Deprizio D, et al. Early
physical and occupational therapy in mechanically ventilated,
critically ill patients: a randomised controlled trial. Lancet 2009;373:
18741882.
34. Burtin C, Clerckx B, Robbeets C, Ferdinande P, Langer D, Troosters T,
Hermans G, Decramer M, Gosselink R. Early exercise in critically
ill patients enhances short-term functional recovery. Crit Care Med
2009;37:24992505.
35. Puthucheary ZA, Rawal J, McPhail M, Connolly B, Ratnayake G, Chan
P, Hopkinson NS, Phadke R, Dew T, Sidhu PS, et al. Acute skeletal
muscle wasting in critical illness. JAMA 2013;310:15911600.
36. Hough CL, Lieu BK, Caldwell ES. Manual muscle strength testing of
critically ill patients: feasibility and interobserver agreement. Crit Care
2011;15:R43.
37. Stuart EA. Matching methods for causal inference: a review and
a look forward. Stat Sci 2010;25:121.
38. Demoule A, Jung B, Prodanovic H, Molinari N, Chanques G, Coirault C,
Matecki S, Duguet A, Similowski T, Jaber S. Diaphragm dysfunction
on admission to the intensive care unit. Prevalence, risk factors,
and prognostic impact. A prospective study. Am J Respir Crit Care
Med 2013;188:213219.
39. Supinski GS, Callahan LA. Diaphragm weakness in mechanically
ventilated critically ill patients. Crit Care 2013;17:R120.
40. de Jonghe B, Lacherade JC, Sharshar T, Outin H. Intensive care unitacquired weakness: risk factors and prevention. Crit Care Med 2009;
37(Suppl. 10):S309S315.
41. Mirzakhani H, Williams JN, Mello J, Joseph S, Meyer MJ, Waak K,
Schmidt U, Kelly E, Eikermann M. Muscle weakness predicts
pharyngeal dysfunction and symptomatic aspiration in long-term
ventilated patients. Anesthesiology 2013;119:389397.
42. Leijten FS, Harinck-de Weerd JE, Poortvliet DC, de Weerd AW. The
role of polyneuropathy in motor convalescence after prolonged
mechanical ventilation. JAMA 1995;274:12211225.
43. Fan E, Dowdy DW, Colantuoni E, Mendez-Tellez PA, Sevransky JE,
Shanholtz C, Himmelfarb CR, Desai SV, Ciesla N, Herridge MS, et al.
Physical complications in acute lung injury survivors: a 2-year
longitudinal prospective study. Crit Care Med 2014;42:849859.

Hermans, Van Mechelen, Clerckx, et al.: Acute Outcomes and 1-Year Mortality of ICUAW

419

ORIGINAL ARTICLE
44. Needham DM, Dinglas VD, Bienvenu OJ, Colantuoni E, Wozniak AW,
Rice TW, Hopkins RO. One year outcomes in patients with acute lung
injury randomised to initial trophic or full enteral feeding: prospective
follow-up of EDEN randomised trial. BMJ 2013;346:f1532.
45. Bagshaw SM, McDermid RC. The role of frailty in outcomes from
critical illness. Curr Opin Crit Care 2013;19:496503.
46. Bagshaw SM, Stelfox HT, McDermid RC, Rolfson DB, Tsuyuki RT, Baig N,
Artiuch B, Ibrahim Q, Stollery DE, Rokosh E, et al. Association between
frailty and short- and long-term outcomes among critically ill patients:
a multicentre prospective cohort study. CMAJ 2014;186:E95E102.
47. Groll DL, To T, Bombardier C, Wright JG. The development of
a comorbidity index with physical function as the outcome. J Clin
Epidemiol 2005;58:595602.

420

48. Gosselink R, Clerckx B, Robbeets C, Vanpee G, Segers J.

Physiotherapy in the intensive care unit. Neth J Crit Care 2010;15:
6675.
49. Gregson JM, Leathley MJ, Moore AP, Smith TL, Sharma AK, Watkins
CL. Reliability of measurements of muscle tone and muscle power in
stroke patients. Age Ageing 2000;29:223228.
50. Kleyweg RP, van der Meche FG, Schmitz PI. Interobserver agreement
in the assessment of muscle strength and functional abilities in
Guillain-Barre syndrome. Muscle Nerve 1991;14:11031109.
51. Fan E, Ciesla ND, Truong AD, Bhoopathi V, Zeger SL, Needham DM.
Inter-rater reliability of manual muscle strength testing in ICU
survivors and simulated patients. Intensive Care Med 2010;36:
10381043.

American Journal of Respiratory and Critical Care Medicine Volume 190 Number 4 | August 15 2014